i18nutil/qa/cppunit/test_scriptchangescanner.cxx | 23 ++++++++++++++++++-----
i18nutil/source/utility/scriptchangescanner.cxx | 23 ++++++++++++++++++-----
2 files changed, 36 insertions(+), 10 deletions(-)
New commits:
commit 9b505f583954c88ce7b72a07c9bfd65d78d863ef
Author: Jonathan Clark <jonat...@libreoffice.org>
AuthorDate: Mon Jan 6 03:36:31 2025 -0700
Commit: Jonathan Clark <jonat...@libreoffice.org>
CommitDate: Mon Jan 6 13:45:19 2025 +0100
tdf#66791 sw: Apply first-seen script type to leading weak characters
Previously, leading weak characters were always assigned a script type
derived from the application language. This change updates the algorithm
to instead use the logical-order first-seen non-weak script type.
Change-Id: I5f6732c6faa1eb35aff51e98e82a87c8008f70ab
Reviewed-on: https://gerrit.libreoffice.org/c/core/+/179824
Tested-by: Jenkins
Reviewed-by: Jonathan Clark <jonat...@libreoffice.org>
diff --git a/i18nutil/qa/cppunit/test_scriptchangescanner.cxx
b/i18nutil/qa/cppunit/test_scriptchangescanner.cxx
index e0726a45d922..1b033d94218a 100644
--- a/i18nutil/qa/cppunit/test_scriptchangescanner.cxx
+++ b/i18nutil/qa/cppunit/test_scriptchangescanner.cxx
@@ -27,6 +27,7 @@ public:
void testTrivial();
void testTrivialAppLang();
void testWeakAtStart();
+ void testOnlyWeak();
void testStrongChange();
void testMongolianAfterNNBSP();
void testNonspacingMark();
@@ -44,6 +45,7 @@ public:
CPPUNIT_TEST(testTrivial);
CPPUNIT_TEST(testTrivialAppLang);
CPPUNIT_TEST(testWeakAtStart);
+ CPPUNIT_TEST(testOnlyWeak);
CPPUNIT_TEST(testStrongChange);
CPPUNIT_TEST(testMongolianAfterNNBSP);
CPPUNIT_TEST(testNonspacingMark);
@@ -101,21 +103,32 @@ void ScriptChangeScannerTest::testTrivialAppLang()
void ScriptChangeScannerTest::testWeakAtStart()
{
+ // The first-seen script type is used for weak characters at the start
auto aText = u"“x”"_ustr;
auto pDirScanner = MakeDirectionChangeScanner(aText, 0);
auto pScanner = MakeScriptChangeScanner(aText,
css::i18n::ScriptType::COMPLEX, *pDirScanner);
CPPUNIT_ASSERT(!pScanner->AtEnd());
- CPPUNIT_ASSERT_EQUAL(css::i18n::ScriptType::COMPLEX,
pScanner->Peek().m_nScriptType);
+ CPPUNIT_ASSERT_EQUAL(css::i18n::ScriptType::LATIN,
pScanner->Peek().m_nScriptType);
CPPUNIT_ASSERT_EQUAL(sal_Int32(0), pScanner->Peek().m_nStartIndex);
- CPPUNIT_ASSERT_EQUAL(sal_Int32(1), pScanner->Peek().m_nEndIndex);
+ CPPUNIT_ASSERT_EQUAL(sal_Int32(3), pScanner->Peek().m_nEndIndex);
pScanner->Advance();
+ CPPUNIT_ASSERT(pScanner->AtEnd());
+}
+
+void ScriptChangeScannerTest::testOnlyWeak()
+{
+ // The application language is used for text containing only weak
characters
+ auto aText = u"“”"_ustr;
+ auto pDirScanner = MakeDirectionChangeScanner(aText, 0);
+ auto pScanner = MakeScriptChangeScanner(aText,
css::i18n::ScriptType::COMPLEX, *pDirScanner);
+
CPPUNIT_ASSERT(!pScanner->AtEnd());
- CPPUNIT_ASSERT_EQUAL(css::i18n::ScriptType::LATIN,
pScanner->Peek().m_nScriptType);
- CPPUNIT_ASSERT_EQUAL(sal_Int32(1), pScanner->Peek().m_nStartIndex);
- CPPUNIT_ASSERT_EQUAL(sal_Int32(3), pScanner->Peek().m_nEndIndex);
+ CPPUNIT_ASSERT_EQUAL(css::i18n::ScriptType::COMPLEX,
pScanner->Peek().m_nScriptType);
+ CPPUNIT_ASSERT_EQUAL(sal_Int32(0), pScanner->Peek().m_nStartIndex);
+ CPPUNIT_ASSERT_EQUAL(sal_Int32(2), pScanner->Peek().m_nEndIndex);
pScanner->Advance();
diff --git a/i18nutil/source/utility/scriptchangescanner.cxx
b/i18nutil/source/utility/scriptchangescanner.cxx
index 2cb8d9d96819..456af39bb495 100644
--- a/i18nutil/source/utility/scriptchangescanner.cxx
+++ b/i18nutil/source/utility/scriptchangescanner.cxx
@@ -132,7 +132,7 @@ private:
ScriptChange m_stCurr;
DirectionChangeScanner* m_pDirScanner;
const OUString& m_rText;
- sal_Int16 m_nPrevScript;
+ sal_Int16 m_nPrevScript = css::i18n::ScriptType::WEAK;
sal_Int32 m_nIndex = 0;
bool m_bAtEnd = false;
bool m_bApplyAsianToWeakQuotes = false;
@@ -142,16 +142,23 @@ public:
DirectionChangeScanner* pDirScanner)
: m_pDirScanner(pDirScanner)
, m_rText(rText)
- , m_nPrevScript(nDefaultScriptType)
{
// tdf#66791: For compatibility with other programs, the Asian script
is
// applied to any weak-script quote characters if the enclosing
paragraph
// contains Chinese- or Japanese-script characters.
+ // In the original Writer algorithm, the application language is used
for
+ // all leading weak characters (#94331#). This implementation deviates
by
+ // instead using the first-seen non-weak script.
sal_Int32 nCjBase = 0;
while (nCjBase < m_rText.getLength())
{
auto nChar = m_rText.iterateCodePoints(&nCjBase);
auto nScript = GetScriptClass(nChar);
+ if (m_nPrevScript == css::i18n::ScriptType::WEAK)
+ {
+ m_nPrevScript = nScript;
+ }
+
if (nScript == css::i18n::ScriptType::COMPLEX)
{
m_bApplyAsianToWeakQuotes = false;
@@ -172,12 +179,18 @@ public:
}
}
- // In the original Writer algorithm, the application language is used
for
- // all leading weak characters. Make a change record for those
characters.
+ // Fall back to the application language for leading weak characters
if a
+ // better candidate was not found.
+ if (m_nPrevScript == css::i18n::ScriptType::WEAK)
+ {
+ m_nPrevScript = nDefaultScriptType;
+ }
+
+ // Make a change record for leading weak characters.
Advance();
if (m_stCurr.m_nStartIndex == m_stCurr.m_nEndIndex)
{
- // The text does not start with application-language leading
characters.
+ // The text does not start with weak characters.
// Initialize with a non-empty record.
Advance();
}
